Offspring of women with gestational diabetes mellitus (GDM) exhibit a high risk of developing type 2 diabetes later in life, potentially as a result of developmental programming by epigenetic mechanisms (1,2). We read with great interest the meta-analysis by Howe et al. (3) integrating results of differential cord blood DNA methylation changes from seven epigenome-wide association studies (EWAS) including in total 3,677 mother-newborn pairs. Pregnancies complicated with GDM were associated with significantly lower cord blood methylation levels within the promoter region of the OR2L13 gene, as well as within the gene body of the CYP2E1 gene (3). At single CpG site levels, the authors reported GDM to be associated with differential methylations at six distinct CpG sites (3). As pointed out by Howe et al., replication of EWAS analyses is extremely important due to 1) high risk of confounders; 2) in general, small absolute methylation differences; and 3) analytical uncertainties related to multiple comparisons and adjustment for cell heterogeneity (3). Accordingly, Howe et al. were unable to replicate any of the previously reported differential DNA methylation differences at distinct CpG sites in cord blood samples from smaller GDM pregnancy cohort studies (3). Another potential major source of noise and errors in epigenetic cord blood studies is the fact that DNA methylations change rapidly at the time of birth and, for instance, have been linked with cesarian section frequently performed in pregnancies affected by GDM (4). Thus, for differential cord blood DNA methylations to be considered significant in a biological and/or clinical biomarker context, stability and consistency over time and age need to be established.
We previously published an epigenome-wide study of 9- to 16-year-old offspring of women with GDM, to date the largest EWAS performed in preadolescent and adolescent GDM offspring (n = 188) (5). Briefly, we measured DNA methylation profiles in peripheral blood from 93 GDM offspring and 95 control subjects using the Illumina HumanMethylation450 BeadChip, and results were analyzed using the two-step RUVm method from the missMethyl Bioconductor package (5). Data of the epigenome-wide array study are available upon reasonable request.
With no need for corrections for multiple comparisons, we reexamined our EWAS data focusing explicitly on the differential methylation results published by Howe et al. (3). We here report that two out of the six differentially methylated CpGs (cg00812770 and cg11187204) found in cord blood from GDM pregnancies by Howe et al. also were significantly or borderline significantly differentially methylated in the same direction at ages 9 to 16 years in the GDM offspring included in our study cohort (Table 1). Furthermore, one CpG site associated with each of the two differentially methylated regions (OR2L13 and CYP2E1) was also differentially methylated in the adolescent GDM offspring at ages 9 to 16 years (Table 1). The other sites reported to be significantly differentially methylated by Howe et al. (cg11723077, cg22791932, cg17588003, and cg10139436) were not significantly different between GDM offspring and control subjects, with nominal P values above 0.10 in our cohort of young offspring.
List of validated CpG sites in the Danish National Birth Cohort, GDM subcohort study (n = 188)
In conclusion, we here uniquely replicate differential DNA methylations reported in cord blood samples from GDM-affected pregnancies in the preadolescent and adolescent offspring of women with GDM. These data support the hypothesis that DNA methylation variations occur early in life in response to GDM and that these changes can be stable and persistent until at least preadolescent and adolescent age. These distinct CpG site methylations may have potential as biomarkers, for instance, to predict risk of developing type 2 diabetes later in life and/or to understand in further detail the mechanisms underlying developmental programming of type 2 diabetes by GDM in pregnancy.
Article Information
Acknowledgments. The authors greatly appreciate all the children and their mothers who participated in the study.
Funding. L.H. is funded by the Danish Diabetes Academy, supported by the Novo Nordisk Foundation and the Danish Diabetes Association.
Duality of Interest. A.V. is shareholder in AstraZeneca. No other potential conflicts of interest relevant to this article were reported.
Author Contributions. L.G.G., S.F.O., and A.V. designed the GDM subcohort study. L.H. processed blood samples, extracted DNA, and performed DNA methylation analyses. L.H. and R.S. analyzed the data, and L.H., L.G.G., R.S., and A.V. interpreted the results of the experiments. L.H. wrote the manuscript with contributions from L.G.G. and A.V. All coauthors revised and approved the manuscript. L.H. and A.V. are guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
- Received January 29, 2021.
- Accepted February 8, 2021.
- © 2021 by the American Diabetes Association